=== harrisj_ is now known as harrisj | ||
pedronis | mvo: hi, you did a full pass on 7020 now? I will address comments today or tomorrow | 07:38 |
---|---|---|
mvo | pedronis: yes | 07:42 |
ackk | hi, we (maas) currently have some scripts that scan /sys to find info about the system (both ardware and configuration/modules/...). this currently doesn't work in the snap, it doesn't seem there is an interfaces that lets you read the whole /sys. Would it be possible to do that? | 09:22 |
sparkiegeek | to add some flavour, and as a non-exhaustive example; we currently have `find /sys/devices/ -name sriov_numvfs` | 09:24 |
ackk | sparkiegeek, that does't actually cause errors AFAICT, the main one that's failing is 'find /sys -name modalias' | 09:28 |
ackk | but also we're using lshw/lscpu/lsusb/... and those don't work | 09:28 |
sparkiegeek | pedronis: ^ thoughts? | 09:49 |
Chipaca | doesn't hardware-observe give you those? | 09:49 |
Chipaca | ackk, sparkiegeek? | 09:50 |
ackk | Chipaca, so, for the find -name modalias I can see the files but the whole /sys is nobody:nogroup so I can't read them as root | 09:50 |
ackk | Chipaca, https://paste.ubuntu.com/p/mkQprpncMQ/ | 09:51 |
ackk | (I assume that's the issue, at least) | 09:51 |
ackk | Chipaca, (that snap does have hardware-observe connected) | 09:52 |
Chipaca | why's it getting nobody'd? | 09:52 |
Chipaca | it's root:root outside | 09:52 |
Chipaca | on 1604 at least | 09:52 |
Chipaca | it's root:root inside as well, in 1604 | 09:52 |
ackk | Chipaca, I wonder if it's because I'm using the daemon-enabled snapd from jdstrand? | 09:54 |
Chipaca | ackk: is it also nobody:nogroup outside on the system you're looking at? | 09:54 |
ackk | Chipaca, oh, it is | 09:54 |
ackk | wth | 09:54 |
ackk | (I'm in a lxd) | 09:54 |
* ackk tries rebooting | 09:55 | |
ackk | nope, also /dev files are nobody:nogroup | 09:55 |
ackk | somethin is messed up here, I guess | 09:56 |
pedronis | Chipaca: thanks for the review | 09:56 |
Chipaca | pedronis: thanks for the code | 09:56 |
Chipaca | pedronis: what do you mean in https://github.com/snapcore/snapd/pull/7083#issuecomment-511885260 ? | 09:56 |
jibel | Hi, I just reported https://forum.snapcraft.io/t/cannot-launch-snaps-on-eoan/12341, is it known? | 09:56 |
pedronis | Chipaca: mvo is looking into that. Now that what we check is the type, we need to make there's only one snap installed with that type (and called snapd) | 09:57 |
pedronis | same logic we have/ad for core | 09:57 |
pedronis | *make sure | 09:57 |
pedronis | *have/had | 09:57 |
Chipaca | easy way would be to enforce the name in install | 09:59 |
Chipaca | ¯\_(ツ)_/¯ | 09:59 |
Chipaca | jibel: I don't know if it's known. I've asked for more info. Please do use preformatted-text for the output of commands (i've edited your post) | 10:00 |
pedronis | Chipaca: that's what the check does | 10:02 |
Chipaca | pedronis: :) | 10:03 |
pedronis | I mean coreNameCheck is that check for core | 10:03 |
jibel | Chipaca, done. | 10:03 |
Chipaca | jibel: something is very broken | 10:04 |
jibel | indeed :) | 10:04 |
Chipaca | jibel: was there a new systemd since you booted last | 10:05 |
Chipaca | I seem to remember somebody saying something about the latest systemd + kernel combo breaking mounts | 10:05 |
Chipaca | not sure that was in eoan though, might've been fedora | 10:05 |
jibel | Chipaca, list of updates before last reboot https://paste.ubuntu.com/p/wSvJmy4r3w/. There is no update of systemd | 10:09 |
Chipaca | jibel: what do you get from zgrep systemd.*running.in.system.mode /var/log/syslog* ? | 10:11 |
pedronis | Chipaca: it was arch, https://forum.snapcraft.io/t/snap-mounts-broken-with-kernel-5-2-and-systemd-242/12272 | 10:11 |
Chipaca | (hoping that still works in eoan) | 10:11 |
Chipaca | things we should add to 'snap version': systemd version, and machine architecture | 10:12 |
jibel | Chipaca, nothing, just very old stuff | 10:13 |
Chipaca | jibel: what systemd version are you on _now_? | 10:13 |
jibel | Chipaca, 240-6ubuntu9 | 10:13 |
Chipaca | jibel: it does look very similar to the thing on arch, but the systemd version is older so i dunno | 10:14 |
Chipaca | maybe the problem is linux 5.2 | 10:14 |
jibel | I can boot on previous kernel | 10:15 |
Chipaca | jibel: systemctl --all --failed |grep snapd | 10:15 |
Chipaca | plz | 10:15 |
jibel | Chipaca, nothing if I grep snapd however if I grep snap I get https://paste.ubuntu.com/p/hdr5qKQrh7/ | 10:16 |
* jibel reboots brb | 10:16 | |
Chipaca | yeah that's bad | 10:17 |
jibel | Chipaca, it works with 5.0 and it's broken with 5.2 | 10:23 |
Chipaca | huzzah | 10:23 |
* Chipaca hugs his 1604 to his chest | 10:24 | |
Chipaca | jibel: please add that to the forum topic | 10:24 |
jibel | I did it | 10:24 |
Chipaca | jibel: awesom | 10:25 |
Chipaca | jibel: I've linked to your topic from the arch one | 10:25 |
Chipaca | jibel: can you work with somebody from the kernel team to bisect this? | 10:25 |
jibel | sure | 10:26 |
* jibel moves -> #ubuntu-kernel | 10:26 | |
Chipaca | jibel: thank you! | 10:29 |
* Chipaca hops on to that # also, out of curiosity | 10:29 | |
* Chipaca breaks a take | 10:40 | |
ogra | sil2100, hey ho, do you know if the CM3+ fix also made it into the classic images (if not, what does it take to update their FW ?) | 11:19 |
sil2100 | ogra: hey, it did, but only for disco and eoan, we didn't SRU it to bionic yet | 11:23 |
ogra | sil2100, i assume thats planned ? | 11:23 |
sil2100 | It might be a good idea, but I don't think we explicitly talked about it - might be that we'd want that before the point release | 11:24 |
sil2100 | I'll poke Dave today | 11:24 |
ogra | thanks ! | 11:24 |
=== ricab is now known as ricab|lunch | ||
ackk | jdstrand, hi, fwiw instructions on https://people.canonical.com/~jamie/snapd-daemon-user/README don't seem to work anymore for me. the test snap doesn't work, but my maas snap does. also the reported version is 2.39.3 | 12:00 |
pedronis | mvo: I did a pass over 7107, there's more work needed | 12:06 |
ackk | Chipaca, fwiw https://paste.ubuntu.com/p/jKyKBmT4Jc/ is what I see | 12:16 |
ackk | that's probably still ok since I think we could just search for modalias under /sys/devices | 12:19 |
ackk | but not entirely sure | 12:19 |
* diddledan pops it in here too for people to have a nosey: https://sc-jsonnet.readthedocs.io/ | 12:40 | |
diddledan | err | 12:40 |
diddledan | wrong url | 12:40 |
diddledan | https://forum.snapcraft.io/t/yaml-code-sharing-and-reuse/12342 | 12:40 |
diddledan | (it was a related url anyway, but I wanted to share my forum post) | 12:40 |
Chipaca | jibel: out of curiosity, what's 'snap list --all | wc -l' for you? | 12:43 |
Chipaca | with 20+ snaps installed, on 5.2.x, with systemd 240, can't repro the issue | 12:44 |
jibel | Chipaca, 22 | 12:55 |
jibel | Chipaca, 41 | 12:55 |
Chipaca | ah | 12:55 |
Chipaca | small difference :) | 12:55 |
jibel | Chipaca, https://paste.ubuntu.com/p/RnkKVNStZX/ | 12:56 |
Chipaca | ok, i'll ramp things up to that level and see | 12:56 |
=== JanC is now known as Guest31310 | ||
Chipaca | yess | 13:12 |
Chipaca | at 42 snaps installed, i got a broken mount | 13:12 |
diddledan | nobody needs more than 42 - that's the perfect number | 13:12 |
* roadmr hands 42 bytes of RAM to diddledan | 13:12 | |
roadmr | ah, snaps - sorry, was missing context :) | 13:12 |
diddledan | hmm | 13:12 |
diddledan | :-p | 13:13 |
roadmr | yes, 42 is ideal | 13:13 |
=== ricab|lunch is now known as ricab | ||
=== JanC_ is now known as JanC | ||
* dot-tobias says hi | 13:28 | |
sparkiegeek | diddledan: https://en.wikipedia.org/wiki/Perfect_number no it's not! :P | 13:32 |
ackk | https://youtu.be/RyFr279K9TE?t=33 | 13:34 |
dot-tobias | zyga: I'm suddenly getting store rejections from failed layout target lintings. The previous revisions went through review just fine, I was mapping $SNAP/some/path to $SNAP_DATA/some/path. Is this expected? | 13:35 |
Chipaca | YESSS i can reproduce the issue without snaps | 13:51 |
Chipaca | wooo woo wo w | 13:51 |
* Chipaca goes to sell it on the black market | 13:52 | |
diddledan | is it a kernel bug? | 13:54 |
Chipaca | no, it's... | 13:55 |
Chipaca | Mail server hit by UniSpammer. | 13:55 |
Chipaca | no wait, it's | 13:55 |
Chipaca | The file system is full of it | 13:55 |
* Chipaca puts away the BOFH snap before it does any real harm | 13:55 | |
mvo | Chipaca: \o/ | 13:56 |
mvo | Chipaca: where can I read more? bugrport? forum? | 13:57 |
Chipaca | mvo: i'm in the progress of writing the reproducer into a script, and then will add it to the forum, and then a bug | 13:57 |
Chipaca | mvo: i'll let you know when it's a script you can run :-) | 13:57 |
mvo | Chipaca: thank you! | 13:58 |
seb128 | mvo, Chipaca, is there any workaround atm? Till was asking because he uses chromium as his browser which was made into a snap by defualt in eoan and he's without a working webbrowser now which is a bit of an issue for work | 14:00 |
jibel | boot with kernel 5.0 | 14:00 |
diddledan | eep | 14:01 |
Chipaca | mvo: https://paste.ubuntu.com/p/VCpzGxvy6h/ | 14:04 |
Chipaca | seb128: snap list --all, for any broken snaps, if disabled remove, if enabled (== current) disable and enable | 14:05 |
Chipaca | seb128: but beware dependencies | 14:05 |
diddledan | worth checking if earlier systems have any problems running that reproducer, too | 14:06 |
Chipaca | seb128: not sure if the bug is in the kernel or in util-linux, but it's in one of them two | 14:06 |
Chipaca | so it could be a lot worse | 14:06 |
dot-tobias | I'm trying to define a content interface for my snaps. Upon installation, I get this error: “INFO snap <snap-name> has bad plugs or slots: my-slot-name (unknown interface "my-slot-name")” I gathered from https://docs.snapcraft.io/content-interface that the slot name can be arbitrary? | 14:09 |
diddledan | dot-tobias: you need to do two things: add a top-level slots: block which defines your slot, and then an app-level slots: block that ties your slot name to an app | 14:10 |
dot-tobias | diddledan: Ok, thanks. Do you happen to have an example how the app-level stanza looks like? Just slots: [my-slot-name]? | 14:14 |
Chipaca | #1836914 | 14:19 |
dot-tobias | diddledan: Hm, followed this example https://github.com/anbox/anbox/blob/cd829e9ccd3a5d654c8aa5e16e32f0d3915d54a8/snap/snapcraft.yaml#L34 but still getting the same error. | 14:19 |
seb128 | Chipaca, thx | 14:27 |
diddledan | dot-tobias: did you also add the top-level item? https://github.com/anbox/anbox/blob/cd829e9ccd3a5d654c8aa5e16e32f0d3915d54a8/snap/snapcraft.yaml#L18 | 14:28 |
ijohnson | is there a clean way to do a type assertion with go-check? I want to check that an error is of a certain type | 14:47 |
diddledan | ijohnson: https://tour.golang.org/methods/15? | 14:48 |
ijohnson | diddledan: specifically with the go-check package: https://godoc.org/gopkg.in/check.v1 | 14:48 |
pedronis | ijohnson: look for FitsTypeOf | 14:48 |
ijohnson | pedronis: ah perfect thanks | 14:49 |
* diddledan goes back to lurkioing | 14:49 | |
* Chipaca would like FitsTypeOf to raise a TooLarge error :-p | 14:49 | |
ijohnson | pedronis: I was going to break out a small PR from netplan apply to make it easy to create hidden commands in snapctl | 14:49 |
diddledan | Chipaca: what about returning Almost? | 14:50 |
pedronis | ijohnson: sounds good | 14:50 |
Chipaca | diddledan: you get it :) | 14:50 |
diddledan | TryHarer? | 14:50 |
diddledan | TryHarder** | 14:50 |
Chipaca | MoreLubricant | 14:50 |
Chipaca | GetAHammer | 14:50 |
diddledan | haha | 14:50 |
pedronis | ijohnson: otherwise of course you can also use a go type assertion, _, ok := x.(T) and check for ok | 14:50 |
pedronis | depends a bit exactly what you need | 14:51 |
diddledan | or IGuessItWorks | 14:51 |
ijohnson | yeah I was hoping there was a go-check thing to use, and there is so that seems more "idiomatic snapd tests" | 14:51 |
pedronis | yes | 14:52 |
ackk | sergiusens, hi, around? | 14:57 |
pedronis | ijohnson: forgot to mention, have you looked how adding hidden commands works in cmd/snap, we probably want the same | 15:03 |
pedronis | or sameish | 15:03 |
ijohnson | pedronis: I did not, is it different than using go-flags Command.Hidden flag? | 15:04 |
pedronis | it's the same | 15:04 |
pedronis | but there's a pattern there | 15:04 |
pedronis | please look :) | 15:04 |
ijohnson | sure, is there a hidden command off the top of your head you could point me to? | 15:04 |
pedronis | ijohnson: interfaces nowadays | 15:05 |
ijohnson | pedronis: ack, this actually looks very similar to what I have now | 15:06 |
sergiusens | ackk: nope, travelling and packing with the family today | 16:18 |
ackk | sergiusens, ah cool, enjoy then | 16:19 |
pedronis | mvo: thanks for adding checkSnapdName. I reviewed 7086 , just one odd thing there (cc Chipaca) | 16:21 |
pedronis | oops, I meant 7084 | 16:21 |
pedronis | heh, fail | 16:21 |
pedronis | mvo: Chipaca: I meant actually really 7083 (3/4) | 16:22 |
mvo | pedronis: no worries, I will check the feedback (either today or tomorrow morning) | 16:22 |
pedronis | thx, I can tweak as well tomorrow if nobody gets to it before | 16:22 |
* pedronis going to eod | 16:22 | |
pedronis | Chipaca: https://forum.snapcraft.io/t/snap-layouts/7207/21 this looks like it should be its own topic | 16:33 |
pedronis | (it's not about the docs but a issue of usage/review tools) | 16:33 |
ijohnson | pedronis: fyi I filed to do hidden snapctl cmds in https://github.com/snapcore/snapd/pull/7119 | 17:12 |
ijohnson | pedronis: I'm now working on another PR for passing through the exit code, etc. from the snapd daemon to the snapctl client as requested | 17:13 |
ijohnson | may or may not get that PR finished today | 17:13 |
pedronis | ijohnson: thx | 17:42 |
jdstrand | ackk: ok, I needed to modify test-snapd-daemon-user and made it not work with the deb I gave you. let me update the deb and README | 17:46 |
ackk | jdstrand, thanks! | 17:52 |
jdstrand | ackk: ok, new deb and updated README are uploaded | 18:13 |
jdstrand | ackk: you'll need to use 'system-usernames: [ snap_daemon ]' in your snap now after all | 18:13 |
jdstrand | ackk: 'system-usernames' is subject to change still (should have a decision tomorrow), but 'snap_daemon' shouldn't change | 18:14 |
jdstrand | (and it may not be a list, but the yaml is the easy part, you can be reasonably confident that changing the rest of your snap to use 'snap_daemon' will mean you're ready when the feature lands | 18:15 |
jdstrand | ) | 18:15 |
jdstrand | roadmr: hey, I know this was a quick one, but can you pull 20190717-1931UTC? | 19:32 |
roadmr | jdstrand: oh hehe :) sure | 19:38 |
jdstrand | roadmr: thanks! :) | 20:38 |
ackk | jdstrand, thanks! | 21:51 |
diddledan | wanna see some voodoo? https://forum.snapcraft.io/t/yaml-code-sharing-and-reuse/12342 | 21:54 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!